The company claims o3 is the most advanced reasoning model it has developed, while o4-mini is said to provide a balance of price, speed, and performance. OpenAI introduced o3 in December, stating that the model could solve approximately 25% of questions on FrontierMath, a difficult math problem set. Questions have been raised regarding OpenAI’s transparency and procedures for testing models after a difference in benchmark outcomes was detected by first- and third-party benchmark results for the o3 AI model. The company skipped that step — sending safety cards for GPT-4.1 — claiming in a statement to TechCrunch that “GPT-4.1 is not a frontier model, so there won’t be a separate system card released for it.” OpenAI leaders have been talking about allowing the open model to link up with OpenAI’s cloud-hosted models to improve its ability to respond to intricate questions, two sources familiar with the situation told TechCrunch. The chatbot provides recommendations, images, and reviews of products in various categories such as fashion, beauty, home goods, and electronics.
OpenAI’s new “company knowledge” update for ChatGPT lets Business, Enterprise, and Education users search workplace data across tools like Slack, Google Drive, and GitHub using GPT‑5, per a report by The Verge. OpenAI launched a new AI shopping feature in ChatGPT ahead of the peak holiday shopping window to help users research potential purchases. On the same day, Disney notably launched a lawsuit against Google alleging “massive” copyright infringement occurring in its AI models. Prior to training, the average quality of the datasets was improved as the model matured from version 1 to version 3.
ChatGPT can directly edit your code
In a post on X, Altman said that, in the coming months, OpenAI will release a model called GPT-5 that “integrates a lot of OpenAI’s technology,” including o3, in ChatGPT and its API. It’s unclear when these agentic tools might launch or which customers will be eligible to buy them. We trained a new model that is good at creative writing (not sure yet how/when it will get released). The company has mostly concentrated on challenges in rigid, predictable areas such as math and programming. He posted a lengthy sample from the model given the prompt “Please write a metafictional literary short story about AI and grief.” OpenAI has not extensively explored the use of AI for writing fiction. OpenAI CEO Sam Altman said, in a post on X, that the company has trained a “new model” that’s “really good” at creative writing.
Altman warns that ChatGPT therapy isn’t confidential
Complex mathematical techniques help the decoder velwins casino to estimate several different outputs and predict the most accurate one. The decoder uses the vector representation to predict the requested output. When it receives input, the transformer network’s encoder block separates words into embeddings and assigns weight to each.
Write and learn code
GPT models generate text in casual, humorous, professional, and other styles. Text-generating AI models like ChatGPT have a tendency to regurgitate content from their training data. OpenAI has said that individuals in “certain jurisdictions” (such as the EU) can object to the processing of their personal information by its AI models by filling out this form. That’s because ChatGPT lacks context awareness — in other words, the generated code isn’t always appropriate for the specific context in which it’s being used. While ChatGPT can write workable Python code, it can’t necessarily program an entire app’s worth of code. And nonprofit organization Solana officially integrated the chatbot into its network with a ChatGPT plug-in geared toward end users to help onboard into the web3 space.
The coding agent may take anywhere from one to 30 minutes to complete tasks such as writing simple features, fixing bugs, answering questions about your codebase, and running tests. OpenAI has introduced its AI coding agent, Codex, powered by codex-1, a version of its o3 AI reasoning model designed for software engineering tasks. The feature enables ChatGPT to retrieve information across users’ own services to answer their questions.
While OpenAI had previously allowed users to ask ChatGPT questions without signing in, responses were restricted to the chatbot’s last training update. ChatGPT users will see an updated “chain of thought” that shows more of the model’s “reasoning” steps and how it arrived at answers to questions. In response to pressure from rivals like DeepSeek, OpenAI is changing the way its o3-mini model communicates its step-by-step “thought” process. Using OpenAI’s latest default model for ChatGPT, GPT-4o, as a reference, nonprofit AI research institute Epoch AI found the average ChatGPT query consumes around 0.3 watt-hours.
OpenAI plans to sunset GPT-4, an AI model introduced more than two years ago, and replace it with GPT-4o, the current default model, per changelog. Developers can use GPT-4.5 through OpenAI’s API until July 14; then, they will need to switch to GPT-4.1, which was released on April 14. However, multiple independent tests indicate that the model is less reliable than previous OpenAI releases. Aidan Clark, OpenAI’s VP of research, is spearheading the development of the open model, which is in the very early stages, sources familiar with the situation told TechCrunch. OpenAI is unveiling a program called OpenAI for Countries, which aims to develop the necessary local infrastructure to serve international AI clients better. The connector will soon be available for ChatGPT Plus, Pro, and Team users, with support for Enterprise and Education coming shortly, per an OpenAI spokesperson.
- OpenAI launched Instant Checkout in ChatGPT, letting U.S. users purchase products directly from Etsy and, soon, over a million Shopify merchants without leaving the conversation.
- We planned to launch our open-weight model next week.we are delaying it; we need time to run additional safety tests and review high-risk areas.
- What started as a tool to supercharge productivity through writing essays and code with short text prompts has evolved into a behemoth with 300 million weekly active users.
- The changes aren’t yet publicly visible, but a user on X who goes by Choi spotted these updates in ChatGPT’s client-side code.
- OpenAI first previewed the model in December alongside a more capable system called o3.
Manage all your machines on the web with Ninite Pro
Also, experienced developers can use GPT tools to autosuggest relevant code snippets. GPT-powered image processing software can create memes, videos, marketing copy, and other content from text instructions. For example, marketers can prompt a GPT model to produce an explainer video script. For example, where you might need several hours to research, write, and edit an article on nuclear physics, a GPT model can produce one in seconds. The value of these models lies in their speed and the scale at which they can operate. The GPT models, and in particular, the transformer architecture that they use, represent a significant AI research breakthrough.
- OpenAI launched a new AI “reasoning” model, o3-mini, the newest in the company’s o family of models.
- OpenAI charges $150 for every million tokens (about 750,000 words) input into the model and $600 for every million tokens the model produces.
- This year alone, the app has generated $1.35 billion, a 673% increase from the same period in 2024, averaging nearly $193 million per month, or 53 times more than its nearest rival, Grok.
- The company skipped that step — sending safety cards for GPT-4.1 — claiming in a statement to TechCrunch that “GPT-4.1 is not a frontier model, so there won’t be a separate system card released for it.”
- The mid-tier plan, which offers higher usage limits, image generation, file uploads, and better memory compared to the free version, enters the market in direct competition with Google’s new AI Plus plan in Indonesia.
- The company has mostly concentrated on challenges in rigid, predictable areas such as math and programming.
- OpenAI made a notable change to its content moderation policies after the success of its new image generator in ChatGPT, which went viral for being able to create Studio Ghibli-style images.
OpenAI is partnering with Walmart to allow users to browse products, plan meals, and make purchases through ChatGPT, with support for third-party sellers expected later this fall. The feature acts as a conversational search engine, providing more comprehensive and accurate answers by scouring multiple sources simultaneously. OpenAI revealed that a small but significant portion of ChatGPT users, more than a million weekly, discuss mental health struggles, including suicidal thoughts, psychosis, or mania, with the AI. OpenAI is exploring the consumer health sector, developing AI tools like personal health assistants and data aggregators, according to a report by Business Insider. A Munich court ruled that ChatGPT violated German copyright law by reproducing lyrics from nine protected songs, including Herbert Grönemeyer’s hits, rejecting OpenAI’s argument that the AI only reflected learned patterns.
By popular request, GPT-4.1 will be available directly in ChatGPT starting today.GPT-4.1 is a specialized model that excels at coding tasks & instruction following. For instance, an analyst could use the company’s slide deck and documents to develop an investment thesis. OpenAI upgraded ChatGPT’s conversational voice mood for all paid users across different markets and platforms.
The partnership will let users create AI videos using hundreds of Disney-owned characters, costumes, and props. The model will roll out to paid ChatGPT users and developers in three versions — Instant, Thinking, and Pro — tailored for everything from everyday tasks to complex reasoning and high-accuracy work. The data underscores OpenAI’s push to win enterprise customers as competition heats up from Google, Anthropic, and open-model rivals, a recurring theme you’ll see in recent updates. OpenAI has added new controls in ChatGPT that let users adjust the chatbot’s warmth, enthusiasm, emoji use, and formatting style.
OpenAI suggests traits like “Chatty,” “Encouraging,” and “Gen Z.” However, some users reported that the new options have disappeared, so it’s possible they went live prematurely. Some users found they can specify a preferred name or nickname and “traits” they’d like the chatbot to have. The feature will start rolling out to ChatGPT Plus, Team, and Pro users around the globe this week.
They analyze natural language queries, known as prompts, and predict the best possible response based on their understanding of language. Though it’s accurate to describe the GPT models as artificial intelligence (AI), this is a broad description. The GPT models allow you to build intelligent interactive voice assistants. The GPT model can help business analysts efficiently compile large volumes of data.
Referrals from ChatGPT to news sites are rising but not enough to offset search declines
OpenAI charges $150 for every million tokens (about 750,000 words) input into the model and $600 for every million tokens the model produces. OpenAI says its o1-pro uses more computing than its o1 “reasoning” AI model to deliver “consistently better responses.” It’s only accessible to select developers who have spent at least $5 on OpenAI API services. The company claims they are improved versions of what was already there and that they hallucinate less. Showing ChatGPT users a tiny disclaimer that the chatbot can make mistakes clearly isn’t enough. Personal data has to be accurate,” said Joakim Söderberg, data protection lawyer at Noyb, in a statement. OpenAI has updated its AI voice assistant with improved chatting capabilities, according to a video posted on Monday (March 24) to the company’s official media channels.
O3-pro is available for ChatGPT and Team users and in the API, while Enterprise and Edu users will get access in the third week of June. The ChatGPT maker is one of the biggest buyers of Nvidia’s GPUs, using the AI chips to train models, and this is the first time that OpenAI is using non-Nvidia chips in an important way. Since Google launched its AI Overviews in May 2024, the percentage of news searches that don’t lead to clicks on news websites has increased from 56% to nearly 69% by May 2025. But this rise is insufficient to offset the decline in clicks as more users now obtain their news directly from AI or AI-powered search results, according to a report by digital market intelligence company Similarweb.